MLJ 2018 - Simple Strategies for Semi-Supervised Feature Selection

Matlab code for the methods presented in:

K. Sechidis, G. Brown, Simple Strategies for Semi-Supervised Feature Selection.
https://link.springer.com/article/10.1007/s10994-017-5648-2

Hypothesis testing in semi-supervised scenarios (Section 3)

Function semiIAMB.m implements our algorithm Semi-IAMB, which is the switching procedure applied to Markov Blanket discovery IAMB (IAMB.m).

Ranking features in semi-supervised scenarios (Section 4)

Functions semiMIM.m and semiJMI.m implement our algorithms Semi-MIM and Semi-JMI, which are the switching procedure applied to the feature selection methods MIM (MIM.m) and JMI (JMI.m) respectively.

Tutorial

The tutorial 'Tutorial_SemiSupervised_FS.m' presents how our suggested methods can be used for feature selection in semi-supervised learning environments.

Citation

If you make use of the code found here, please cite the paper above.

@article{sechidis2017semisupervised,
title = {Simple strategies for semi-supervised feature selection},
author = {Konstantinos Sechidis and Gavin Brown},
journal = {Machine Learning},
volume = {107},
number = {2},
pages = {357--395},
year = 2018
}